nsqadmin: timeout settings

[mschneider82]

Hi,

I justed upgraded my infrastructure (~120 nsqd's, ~20 topics, serval channels) from 0.2.9 to 0.3.8 and now the nsqadmin is always failing because of timeout settings:

Example1:

clicking on a topic: http://nsqadmin.local/topics/exampletopic

ERROR: failed to decode JSON - Unexpected token in JSON at position 132

Browser inspect:

Headers:

Request URL:http://nsqadmin.local/api/topics/exampletopic
Request Method:GET
Status Code:502 Bad Gateway

Response:

{"message":"UPSTREAM_ERROR: Failed to query any nsqlookupd: Get http://nsqlookupd02.local:4161/lookup?topic=exampletopic: dial tcp: i/o timeout
Get http://nsqlookupd01.local:4161/lookup?topic=exampletopic: dial tcp: i/o timeout"}

this also happens when i browse to /nodes, I have about ~120 nsqd listed some may be on high load servers.

Example2:

sometimes when i click on a topic it works but not all nsqd's are listed, a WARNING is displayed at the top that some of the servers cannot be reached:

WARNING: Get http://server1.local:4151/stats?format=json: dial tcp: i/o timeout Get http://server2.local:4151/stats?format=json: dial tcp: i/o timeout Get 
...20 more servers...

i manuelly checked all of them all nsqd's are working fine, also the json response is OK
This issue also happens in v0.3.7

[mschneider82]

i just saw 454cf2f

is http-client-connect-timeout also a commandline parameter? I cant find it with --help

[jehiah]

FWIW display of errors in the nsqadmin UI when it fails to connect to nsqlookupd or nsqd is new as of #657 (0.3.6) so versions before that may have allowed errors to passed undetected.

As of #776 nsqadmin does have -http-client-connect-timeout duration and -http-client-request-timeout duration. You'll need to build a copy of nsqadmin from master to get that.

[mreiferson]

Yea, probably time for a release?

[mschneider82]

i compiled it and its working with a 20s timeout setting 👍
a new stable release would be great. maybe enlarge the default parameters?